Valdris Agent Knowledge Vault

This is the agent-facing OKF bundle for the Valdris SDLC Harness. Start here, choose the smallest relevant section, then open the linked concept or playbook.

Systems

• Valdris SDLC Harness - the repository-level SDLC assurance harness this repo implements around external coding agents.
• Connector Event Runtime - bridge, events, artifacts, approvals, and finish-line enforcement.

Playbooks

• Engineering Task Routing - use for normal Codex engineering tasks.
• GitNexus Code Intelligence - use before codebase, architecture, debugging, refactor, or cross-file claims.
• Layer Zero Assurance - bind workload classification to foundation capabilities, tiers, profiles, and proof before delivery.
• Production Readiness 13 Domains - use when a task can affect production behavior.
• Goal Loop and Skill Routing - turn a request into one primary skill, bounded checkpoints, and proof-bearing stopping conditions.
• Generative AI Assurance - use when models, prompts, retrieval, tools, memory, or agents affect behavior.
• Clean-Room Assurance Import - port reusable assurance behavior without importing private project identity, evidence, or topology.
• Semantic and Authoritative Assurance - separate structural, semantic, and externally authoritative claims with typed operating contracts and v3 packet closure.

Concepts

• Proof-First Harness - the core operating principle behind gates and final answers.
• OKF Agent Vault - why this knowledge/ folder exists and how agents should maintain it.
• Typed Evidence - the control-level proof contract for enterprise and AI gates.
• Ontology-Grounded Classification - classify from mechanism, criteria, and evidence; use authoritative web sources when local support is incomplete.

Sources

• OKF and LLM-Wiki Source Notes - source references behind the vault shape.